Wavelet-Based Multi-scale GVF Snake Model for Image Segmentation

Gradient Vector Flow snake creates its own external GVF force field, this makes itself has the ability to move into boundary concavities and can be initialized far away from the boundary. However, the cost of the above advancement is the larger amount of computation and the higher sensitivity to noise. After wavelet transform, the local module maxima of the image´s wavelet coefficients vary in different way in multi resolution due to the different singularities of signal and noise, so noise can also be distinguished from signal with multi-scale GVF snake model. In the lower resolution, there are less wavelet coefficients and the GVF snake is easy to deform to the contour without much computation and is less interfered by noise. In higher resolution, on the basis of the initial position of the foregoing resolution, much more computation will be saved. The 3-order spline function has similar structure with Gauss function as the smoothing function, its derivative is a compact support function that could be used as B-spline wavelet. From some experiments, it can be seen that the multi-scale GVF snake model is more quickly and robust contrast to GVF snake model.